PM Portfolio

Bus ETA System

"Real-Time Bus Arrival Intelligence for Hong Kong"

A city-scale ETA platform combining GPS tracking, ML-powered timetabling, and multi-channel dissemination to deliver accurate bus arrival predictions across Hong Kong's entire franchised bus network.

3,800+
Buses Tracked
380+
Routes Covered
80%
Accuracy Improvement
25%
Satisfaction Boost

Hong Kong's Bus Network Needs Real-Time Intelligence

Aging timetabling systems, GPS interference from dense urban terrain, and static schedules that can't keep up with dynamic traffic conditions.

🕑

Outdated Timetabling

Bus timetabling systems over a decade old could not capture real-time operational constraints. Manually scheduling one route took an entire day, leading to fleet-wide inefficiencies.

📡

GPS Disruption in Dense Urban Environment

Hong Kong's tightly packed high-rises create severe urban canyon effects, while glass curtain walls reflect GPS signals causing positioning errors. Tunnels force complete signal blackouts.

📊

Hidden Speed Limit Data

Until mid-2018, bus operators were unaware that the Transport Department provided free digital speed limit maps, preventing accurate speeding monitoring across all road sections.

🕒

Dynamic Traffic Outpacing Static Schedules

Deteriorating traffic congestion frequently caused actual journey times to exceed government-published schedules. Older systems couldn't recalculate ETAs in real-time.

Who We're Building For

Primary User

The Daily Commuter

Millions of Hong Kong residents relying on buses as their primary transport. Needs precise arrival information to plan journeys and minimize waiting time.

  • Get instant access to real-time bus ETAs via mobile app
  • View countdown timers and "Scheduled Bus" status at a glance
  • Receive alight reminders before their destination stop
Secondary User

The Bus Captain

One of 7,500+ frontline drivers serving 2.7M passenger trips daily. Regulated shifts (max 12-hour duty, 10-hour driving) with mandatory 40-minute rest after 6 hours.

  • Receive ML-optimized duty rosters respecting rest and driving limits
  • Get real-time speeding and harsh braking alerts from black boxes
  • Balanced shift assignments across peak and off-peak periods
Extended User

The Fleet Manager

Operations leadership deploying 3,800+ buses across 380+ routes. Requires high-speed scenario comparison to optimize fleet utilization.

  • Compare operational scenarios and optimize bus deployment at scale
  • Monitor actual vs. scheduled departures for regulatory compliance
  • Generate timetables in seconds, not days

From App to Arrival

1

Pre-Trip Planning

Open the mobile app to find the nearest stop and check the ETA. See countdown or "Scheduled Bus" status.

⚠ GPS interference from high-rises degrades accuracy
2

En-Route to Stop

Walk to the bus stop guided by the countdown timer, adjusting pace based on live updates.

⚠ Traffic congestion shifts ETA in final minutes
3

At the Bus Stop

IBSID panels at major interchanges and powered stops show visual arrival times and traffic conditions.

⚠ Not all stops have power for display panels
4

Onboard the Bus

"Alight Reminder" alerts two stops before destination. LED announcements in Cantonese, English, and Putonghua.

⚠ GPS drift in tunnels affects announcements

What Users Need to Accomplish

As a daily commuter
I want to see a real-time countdown to the next bus on my phone
So that I can time my walk to the stop and avoid unnecessary waiting.
As a visually impaired passenger
I need automated voice announcements in Cantonese, English, and Putonghua
So that I know exactly when to get off the bus.
As a bus captain
I want ML-optimized duty rosters that respect my rest and driving hour limits
So that I can avoid fatigue and maintain safe service standards.
As a fleet manager
I want to generate optimized timetables and duty assignments in under 30 seconds
So that I don't spend an entire day manually scheduling one route.

City-Scale Real-Time Intelligence

CORE
📡

Real-Time GPS Tracking

Onboard units transmit vehicle coordinates and status every 30 seconds. Dead Reckoning and Snap-on-route correct for signal loss in tunnels and urban canyons.

✓ 3,800+ buses across 380+ routes
CORE
📱

Multi-Channel ETA Dissemination

Arrival information shared via mobile app, home screen widgets, and roadside IBSID panels at major interchanges and powered bus stops.

✓ 2.7 million passenger trips served daily
CORE
🚚

Safety Telematics

Onboard black boxes provide real-time audio/visual alerts to bus captains for speeding and harsh braking, with data fed back to fleet management.

✓ 80% improvement in ETA accuracy
CORE
🌐

Snap-on-Route Correction

Matches GPS coordinates to digital maps when signals are disrupted by glass curtain wall reflections and urban canyon effects.

✓ Continuous updates in signal-degraded zones

System Design & Data Flow

Edge Layer
Bach/Mozart Boxes
1s Sampling
GPS Transmitter
30s Intervals
Black Box
Safety Telematics

Wireless Data Transmission
Processing Layer
Server Cluster
Constraint Programming
ETA Engine
Traffic + Historical
Dead Reckoning
Tunnel Positioning

Calculated ETAs & Positioning
Integration Layer
FMI System
Fleet Management
Snap-on-Route
Digital Maps
OCM System
Incident Feed

Enriched ETA Data
Distribution Layer
JSON API
data.etabus.gov.hk
Mobile App
Passenger ETA
IBSID Panels
Roadside Display
Constraint Programming Genetic Algorithms GPS/GNSS Dead Reckoning Snap-on-Route ZF Openmatics JSON REST API Digital Mapping Real-Time Telematics

🚀 Scale

3,800+ buses, 380+ routes, 2.7 million daily passenger trips — all served by a single integrated real-time platform.

📊 Resilience

Multi-layered positioning fallbacks (GPS, Dead Reckoning, Snap-on-Route) ensure continuous ETA accuracy even in Hong Kong's most signal-hostile corridors.

⚙ Feedback Loop

The OCM incident system feeds accidents, typhoons, and disruptions back into ETA calculations in real-time, keeping predictions trustworthy.